Strategies for the design of ligands that combine with metal ions to form high-symmetry coordination assemblies are reviewed. Evaluation of crystal structure evidence reveals that prior design approaches, based on the concept of complementary bonding vector angles, fail to predict the majority of known examples. After explaining the reasons for this failure, it is shown how an alternative approach, de novo structure-based design, provides a practical method that predicts a much wider range of component shapes encoded to direct the formation of such assemblies.